In the last 100 years, scientists have uncovered and developed new tools and methods that offer the ability to peer into the inner layers of the Earth and learn more about the location of valuable minerals, geothermal hotspots, and reservoirs, which house carbon dioxide.
A significant amount of data is collected by observing seismic waves that tend to be produced by earthquakes or artificial events like the detonation of explosives and the use of underwater air guns. How these waves act as they travel through the rock layer provides the necessary feedback.
A minimal range of seismic waves that tack place at frequencies around one hertz could allow researchers to learn more about many of the underground structures that could be encountered. However, these waves also tend to be drowned by the seismic hum of the planet, which makes them hard to track down with the help of conventional detectors. The generation of specific frequencies of this type would require a significant amount of energy.
Complex machine learning system found some mysterious vibrations in earthquake data
A team of researchers has elaborated a new method that could be used to track down the elusive waves. They loaded hundreds of earthquake simulations into a neural network and trained it to recognize the missing frequencies in the case of high-frequency seismic events.
With the help of the new method, researchers may gain the ability to generate the low-frequency waves, which tend to vanish from seismic data. The waves will then be used to map the internal structure of the Earth.
One of the researchers who contributed to the project has stated that the goal of the project is to pave the way towards the mapping of the entire subsurface, with the ability to trace down specific targets without problems. The use of in-depth learning solutions is quite promising at this stage, and the technology could be perfected in the future. A paper was published in a scientific journal.
